Analyzing Scientific Corpora Using Word Embedding

Abstract:

The bibliographic databases have abstract and citations of scientific articles, the summary being the most consulted section of an article. In order to classify and address the entries in a system of indexing and retrieval of information in the databases of a manuscript, there are keywords, which in many cases this information should not achieve greater dissemination. This paper presents an evaluation of the semantic relatedness between the abstract of scientific papers and their keywords. This analysis will be using word2vec that is a pbkp_redictive model, and it will find the nearest words. Thus, this study is focused on the metadata quality assessment through the similar semantics between two words that allow the accuracy in relation to metadata of scientific databases.

Año de publicación:

2019

Keywords:

• Word2Vec
• Natural Language processing
• accuracy
• word embedding

Fuente:

google

scopus